First proposed by March (1991, 1995), the theory of organizational learning describes two distinct and complementary ways in which organizations learn: exploration and exploitation. Exploration initiatives are associated with activities that increase variation in organizational processes, functions, and tasks, including invention, tight control, risk taking, and solutions for new value propositions for customers, which comprise organizational aspects of search, discovery, experimentation, and risk taking (Gupta, Smith, & Shalley, 2006; Popadiuk, 2012; Scandelari & Cunha, 2013).

Thus, exploration activities involve experimenting with new ideas, paradigms, technologies, strategies, and knowledge to discover alternatives that can overcome or at least meet the needs of the market (Jansen et al., 2009; Lewin & Volberda, 1999; Xue, Ray, & Sambamurthy, 2012). Companies that position themselves with exploration innovation practices develop a capacity to frequently map the general external environment and identify factors that enhance the launch of new products and services, differentiate themselves from competitors, and establish themselves as a vanguard (Chen, Mocker, Preston, & Teubner, 2010; Ho & Lu, 2015; Mintzberg, Ahlstrand, & Lampel, 2009).

Exploitation strategies are related to innovation through the use of resources, processes, and strategies of incremental innovation (March, 1991; Scandelari & Cunha, 2013) and are designed to meet the needs of current customers and markets (Ho & Lu, 2015; Maletič, Maletič, & Gomišček, 2016; Popadiuk et al., 2010). Thus, the essence of exploitation innovation seeks to continuously improve existing skills, technologies, and paradigms (Gupta et al., 2006; Jansen et al., 2006). According to Jansen, Van Den Bosch and Volberda (2006), exploitation innovation involves improving existing products and services with frequent and minor adaptations made to the portfolio to maintain and/or expand participation in the current customer market.

Companies that adopt exploitation innovation practices develop skills that help them increase their efficiency and productivity by rationalizing the use of resources and by innovating existing products and services (Popadiuk, 2012). Exploitation innovation is characterized by risk aversion; seeking continuous improvements through existing capacities, skills, and technologies in the rationalization of business processes (Lewin & Volberda, 1999; Popadiuk & Bido, 2016), legitimizing standardization; and automating routines with a strong appeal to productivity in generating economies of scale (Gupta et al., 2006; Xue et al., 2012).

Typically, exploration innovations (or radical innovations) are focused on reaching customers or emerging markets and soliciting new organizational knowledge in contrast to exploitation innovations (or incremental innovations), which are designed to meet the needs of existing customers based on organizational knowledge (Benner & Tushman, 2003).

Seminal academic studies have shown that SIS enable strategies (Chan & Huff, 1992; King, 1978); support business strategy processes and content (Arvidsson, Holmström, & Lyytinen, 2014; Chen et al., 2010; Newkirk & Lederer, 2006; Singh, Watson, & Watson, 2002); and contribute to the survival, support, and growth of organizations (Chan, Sabherwal, & Thatcher, 2006; Chen et al., 2014; Marabelli & Galliers, 2017) in complex environments (Merali et al., 2012) and dynamics (Neirotti & Raguseo, 2017).

The term SIS is treated in the academic literature on IS with several theoretical approaches (Chen et al., 2010; Merali et al., 2012), without having a single definition (Martinez-Simarro, Devece, & Llopis-Albert, 2015; Peppard, Galliers, & Thorogood, 2014). SIS can be seen as a set of IT/IS resources covering the collection, storage, processing, analysis, and availability of data/information to support decision-making and strategic management processes (Chan et al., 2006; Yoshikuni & Albertin, 2018).

The definition of SIS used in this article is provided in a recent study by Yoshikuni and Albertin (2018) and highlighted by Kaplan and Norton (2008) as the IT/IS resources incorporated into the strategic planning process in the phases of strategic awareness, situation analysis, strategy design, formulation, implementation, and business strategy monitoring. Other studies on SIS have investigated their abilities to enable innovation initiatives (Chan et al., 2006; Chen et al., 2010; Johnson & Lederer, 2013; Leidner, Lo, & Preston, 2011; Sabherwal & Chan, 2001) and influence external environmental dynamism (Merali et al., 2012; Mikalef & Pateli, 2017).

The literature on SIS (Merali et al., 2012; Nan & Tanriverdi, 2017; Pavlou & El Sawy, 2010) reiterates the need to direct research toward addressing increased turbulence, uncertainty, and dynamism found in the competitive scenario. Although the importance of SIS to innovation has often been highlighted (Chen et al., 2010; Chuang & Lin, 2017; Leidner et al., 2011; Sabherwal & Chan, 2001), gaps have been identified, necessitating studies using an exploration and exploitation innovation approach enabled by SIS accounting for the influence of environmental dynamism (Pavlou & El Sawy, 2010; Schilke, 2014).

The conceptual framework proposed by Galliers (2011) evidences theoretical ways of understanding relationships between SIS and exploration and exploitation innovation designed to support communication, collaboration, and the leveraging of knowledge processes through related organizational learning strategies in the midst of dynamism in the environment. However, no further studies (on the SIS) have analyzed this relationship and its effects using the conceptual framework proposed by Galliers (2011). A recent IS study (Teubner, 2013) highlights that organizations must react to changes imposed by uncertainty in the environment in an orchestrated and organized way, presenting SIS as an instrument of strategic knowledge management and organizational learning.

Additional studies demonstrate that SIS provide essential means for an organization to effectively develop creative and/or productivity/control initiatives (Chen et al., 2010; Merali et al., 2012). Exploration innovation focuses on a company’s creativity through the generation of new products and services and through new approaches supported by SIS (Leidner et al., 2011) while exploitation innovation is enabled by SIS to develop capabilities with a focus on control or organizational efficiency and productivity (Philip, 2007; Xue et al., 2012).

According to Teubner (2013), SIS contribute to exploration innovation by enabling informal and creative processes of organizational strategies. Above all, these coordination and learning processes take place in individual planning teams at various levels of the organization, from senior management to project committees. Thus, SIS enable the communication, integration, and cooperation of ‘top-down’ and ‘bottom-up’ strategic initiatives, supporting target objective agreements at different levels of the company (Chen et al., 2010; Kaplan & Norton, 2008).

The activities of exploration innovation are fueled by information captured in the external environment (Jansen et al., 2006). The SIS, through digital technologies, identify, collect, process, and analyze a large volume of data and support a company in developing cutting-edge strategies (Davenport, Harris, & Morison, 2010; Jarzabkowski & Kaplan, 2015; Yoshikuni & Albertin, 2018). Thus, SIS support exploration initiatives by providing information for analyzing the life cycles of products and services and simulating the absolute nature of the portfolio and the product/service rupture curve (Merali et al., 2012).

Exploitation innovation is characterized by deliberate analyses oriented toward an objective and partially formalized by decision processes (Teubner, 2013). Exploitation initiatives are supported by IT/IS applications and predetermined operational objectives (Philip, 2007). In other words, exploitation initiatives are focused on the efficient and effective performance of activities that contribute to organizational productivity through incremental innovations (Jansen et al., 2006). Thus, control and monitoring activities are enabled by SIS and contribute to exploitation initiatives (Merali et al., 2012; Yoshikuni & Albertin, 2018) through performance management system applications that consolidate and integrate data and information to measure the effectiveness of planned versus performed activities (Kaplan & Norton, 2008).

In this way, SIS enable effective business strategies to create value and benefits for the process and content of exploitation innovation strategies (by developing strategies to defend the conquered market based on operational efficiency and incremental improvements in products/services) (Philip, 2007; Yoshikuni & Albertin, 2018) and for exploration innovation (by enabling the organization to understand and meet market changes) (Wilden & Gudergan, 2014) and develop proactive strategies (prospecting), supporting strategic decision-making in an agile and effective way (Chan et al., 2006; Sabherwal & Chan, 2001; Xue et al., 2012).

SIS enable flexibility and agility during the formulation of strategic planning and implementation of business strategies (Kearns & Sabherwal, 2006; Yoshikuni & Albertin, 2018) and exploration and exploitation innovation (Johnson & Lederer, 2013; Marabelli & Galliers, 2017; Merali et al., 2012). Through the lens provided by Chan and Huff (1992) and Johnson and Lederer (2013), SIS enable different strategic postures such as aggressiveness, analysis, internal and external defensiveness, externalization, futuristic planning, proactivity, risk taking, and innovation. The various strategic postures described (Chan et al., 2006) reflect exploration and exploitation innovation actions.

Thus, our hypotheses state that SIS (incorporated into business strategies) influence exploration and exploitation innovation in an organization.

H1a: SIS positively influence exploration innovation.

H1b: SIS positively influence exploitation innovation.

Environmental dynamism has been widely studied as a factor that challenges organizations to respond quickly and flexibly to the needs of the external environment (Barbero, Ramos, & Chiang, 2017; Chen et al., 2017; Schilke, 2014). IS researchers also highlight contextual factors that influence the relationship between SIS and organizational effectiveness (Merali et al., 2012; Newkirk & Lederer, 2010; Ray, Wu, & Konana, 2009; Sohn, You, Lee, & Lee, 2003; Yayla & Hu, 2012). The breadth and changes imposed by competition, technologies, new consumer habits, volatility, and instability in the external environment are dimensions that characterize environmental dynamism (Jansen et al., 2006; Kamasak et al., 2017; Wilhelm, Schlömer, & Maurer, 2015).

Dynamism is defined by the level of turbulence or instability faced in an environment that provides substantial evidence of its effects on an organization’s performance (Barbero et al., 2017). Environmental dynamism is defined by the rate and unpredictability of change in a company’s external environment and is characterized by changes in technologies, variations in customer preferences, and fluctuations in product demand or material supply (Jansen et al., 2006, 2009).

Dynamism, as a factor of external volatility, places pressure on an organization to obtain information more quickly and to then understand and make decisions in a constantly changing environment (Chen et al., 2014; Mao, Liu, & Zhang, 2014). Technology enables the organization to generate new knowledge and identify new opportunities by capturing market information (Dameron, Lê, & Lebaron, 2015; George, Haas, & Pentland, 2014; Yoshikuni & Albertin, 2017), analyzing a broad and large volume of data (Rouhani, Ashrafi, Ravasan, & Afshari, 2016; Shollo & Galliers, 2016), transferring data from customers and competitors, and generating rapidly available information (Chen et al., 2014) for agile and flexible decision-making (Pavlou & El Sawy, 2006, 2010) and for formulating and implementing exploration and exploitation initiatives (Mikalef & Pateli, 2017).

From evidence of the presence of different levels of dynamism (high and low) (Chen et al., 2017) reflecting different moderation effects on innovation initiatives (Jansen et al., 2009) and requiring more organizational capacity to face the challenges of environmental uncertainty (Chen et al., 2014; Mao et al., 2014), and based on studies on models of the strategic alignment of technology demonstrating effects on the presence (high and low) of environmental dynamism (Mikalef & Pateli, 2017; Newkirk & Lederer, 2006, 2010; Yayla & Hu, 2012), two additional hypotheses were formulated.

H2a: The positive influence of SIS and exploration innovation will be strongest when environmental dynamism is high.

H2b: The positive influence of SIS and exploitation innovation will be strongest when environmental dynamism is high.

An organization’s decision-making in the face of high dynamism can be supported by SIS adopting a ‘top-down’ approach emphasizing exploitation initiatives (e.g., through steering committees of innovation projects) (Arend, Zhao, Song, & Im, 2017) and emerging initiatives characterized by the coordination and learning of teams and individuals at various organization levels via a ‘bottom-up’ process (Teubner, 2013).

The data used to test the research hypotheses were collected from Brazilian companies using a research instrument. The companies were selected from the directory provided by the Center for Applied Information Technology (FGVcia) of Fundação Getulio Vargas (FGV), as this center’s respondents are managers and executives representing organizations of different sectors and sizes with knowledge of business management practices and technology, thus allowing us to examine our hypotheses. Additionally, the respondents were aware of the FGVcia’s mission, which is to stimulate and coordinate research efforts in information technology to synergistically contribute to the generation of academic and business knowledge.

We sent 1,353 invitations via email to the organizations, and data collection was carried out using an online form (survey) available on the internet. Individuals’ positions, experience, and knowledge regarding the content of the model’s constructs were taken into account when choosing respondents. The sample of respondents obtained covers 256 organizations (19% of the invitations sent), making it a convenience sample (Etikan, Musa, & Alkassim, 2016) or nonprobabilistic or random sample of organizations from a population easily accessed by researchers. Several other studies conducted on IS (Chen et al., 2014; Gupta et al., 2006; Mikalef & Pateli, 2017) have also used samples of convenience for empirical validation.

No missing values were identified, and no errors in questionnaire completion occurred since the internet platform used was configured to restrict errors. The existence of outliers was analyzed by the Mahalanobis square distance (DM²) (Cousineau & Chartier, 2017; Marôco, 2010) using SPSS software, and the sample did not show high DM² values (maximum statistical residual of the DM² variable = 11.756 with a probability p-value < 0.001), indicating an absence of multivariate outliers.

We used the organization as our research analysis unit, and the respondent sample includes business and IT/IS executives who are superintendents, presidents, or directors (39%); managers and coordinators (36%); and supervisors with decision-making power (25%). Table 2 describes the composition of the sampled organizations by sector and the number of employees. The sample is mainly composed of organizations from the service and industry sectors (93% of the companies surveyed) and 40% of the organizations have 500 or more employees.

Table 2
Sample demographics - sector and number of employees.

Note. Source: Elaborated by the authors.

The hypotheses were tested using partial least squares path modeling (PLS-PM) and SmartPLS v3 software (Hair, Hult, Ringle, & Sarstedt, 2017). This approach proved adequate in testing the relationships between the latent variables, analyzing collection bias through formative indicators of the MLMV variable, analyzing the moderation of the continuous variable, comparing differences in path effects of the different groups of the same sample (PLS-MGA), and identifying heterogeneity not observed through PLS-FIMIX and PLS-POS.

In the social sciences in general, the statistical test of ‘power’ should result in a value > 0.8 (Cohen, 1988), meaning that there is at least 80% chance of finding relationships that exist (Goodhue, Lewis, & Thompson, 2007). Researchers Peng and Lai suggest “a post hoc analysis of ‘power’ should be conducted to verify suitability for the study” (Peng & Lai, 2012, p. 47). Following from this, Aguirre-Urreta and Rönkkö (2015) recommended that methodological practices for evaluating ‘power’ in studies using PLS could be improved with the inclusion of a ‘power’ analysis report. We used Aguirre-Urreta and Rönkkö (2015), approach and found consistent results for the sample of 256 cases and a ‘power’ value > 0.8 from the model relationship test.

The latent reflexive variables were subjected to tests of reliability, convergent validity, and discriminant validity. The estimated coefficients (outer loading) of the items show statistical significance (p-value < 0.001) and reported values above 0.7, and items with values between 0.4 and 0.7 were maintained to increase the composite reliability (CR) and average variance extracted (AVE) as recommended by Hair, Sarstedt, Matthews and Ringle (2016) and Hair et al. (2016), confirming convergent validity (see Appendix A). All constructs show average extracted variance (AVE) values above 0.5 (Fornell & Larcker, 1981).

The reliability of a model can be assessed by Cronbach’s alpha or by composite reliability, and in the context of modeling structural equations and PLS-PM, Cronbach’s alpha is sensitive to the number of items in a variable with reliability being considered more appropriate according to Hair et al. (2016). The reliability of the constructs proved to be adequate, with values of composite reliability (CR) exceeding the limit of 0.60 (Chin, 1998; Hair et al., 2017; Henseler, Ringle, & Sinkovics, 2009). Table 3 shows that values on the diagonal (square root of the extracted average variance) are higher than those outside the diagonal (correlations), demonstrating the existence of discriminant validity (Hair et al., 2017; Ringle, Bido, & Silva, 2014). The results of the HTMT test (heterotrait-monotrait ratio) show values below 0.85, confirming discriminant validity as recommended by Henseler, Ringle, and Sarstedt (2015).

Table 3
Correlation matrix between constructs and other measurements.

Note. Diagonal values denote the square root of the AVE (first block). Desirable ranges of metrics indicative of model quality = AVE > 0.5 (Fornell & Larcker, 1981); CA and CC range from 0.6 to 0.9 (Chin, 1998; Hair et al., 2017; Henseler et al., 2009). Source: Elaborated by the authors.

The operationalization of the model involved the assessment of the moderating effect of environmental dynamism as well as the inclusion of control variables (company and sector size); see Table 4. To measure the size of the effect (f²), the methodological literature’s recommendation was followed (Cohen, 1988; Hair, Sarstedt, Ringle, & Gudergan, 2018; Henseler et al., 2009), which suggests that effect size values of 0.02, 0.15, and 0.35 are small, moderate, and large, respectively.

Table 4
Standardized regression coefficients of structural models with all variables.

Note. Sectors were measured by two formative indicators (dummy) to represent the following categories: Agribusiness, Government, Industry, and Services. Significance was estimated by bootstrapping with n = 256 cases and 5,000 repetitions in SmartPLS v3. Bias collection measurement was performed with the inclusion of MLMV indicators. Desirable ranges of metrics indicative of model quality = f² effect size of 0.02 (small), 0.15 (moderate), or 0.35 (large) (Cohen, 1988; Henseler et al., 2009); p-value < 0.05 (Hair et al., 2017); R² values of 0.67 (substantial), 0.33 (moderate), or 0.19 (weak) (Chin, 1998; Henseler et al., 2009). Source: Elaborated by the authors.

The possibility of heterogeneity in the sample (composed of companies from different sectors and of different sizes) was used as a justification for the inclusion of the sector and size control variables (number of employees) in the model studied. The information systems literature includes studies (Benitez, Castillo, Llorens, & Braojos, 2018; Benitez, Llorens, & Braojos, 2018) on the impacts of information technology (IT) on innovation using the PLS-PM statistical technique and using sector and size control variables. In the research results, only the sector control variable of the dependent variable of exploration innovation shows statistical significance (p < 0.01) with a negative path coefficient.

Hypothesis H1a is supported (β = 0.455; f² = 0.331; p-value < 0.001; R² = 51.7%) confirming a large and significant SIS effect of exploration innovation. Hypothesis H1b is also supported (β = 0.550; f² = 0.484; p-value < 0.001; R² = 51.5%) confirming that SIS influence exploitation innovation.

Differences in path effects of the relationships between SIS and exploration innovation and SIS and exploitation innovation were compared, and a difference of 0.095 in the structural coefficients was found.

To verify the existence of statistical significance in this difference between path coefficients of the same sample, we used the methodological framework to test differences between path coefficients following Rodríguez-Entrena, Schuberth and Gelhard (2018). Statistical significance was estimated using the bootstrap technique with 5,000 subsamples and using a 95% confidence interval. Standard/Student’s t confidence interval, percentile bootstrap, and standard bootstrap tests verify that values between the confidence intervals are less than zero, demonstrating a lack of statistical significance (p-value > 0.05) for a difference of 0.095 (17%) between path coefficients in the relationships of SIS -> exploration and SIS -> exploitation (see Table 5).

Table 5
Statistical significance test of two estimates of path coefficients by PLS-PM.

Note. Source: Elaborated by the authors.

Hypothesis H2b is not supported (β = 0.009; f² = 0.00; p-value > 0.05; R² = 51.5%), as the continuous dynamism variable does not show a statistically significant moderating effect on the relationship between SIS and exploitation innovation. However, for the relationship between SIS and exploration innovation, the dynamism variable shows a positive relationship and a large effect with statistical significance (β = 0.135; f² = 0.043; p-value < 0.001; R² = 51.7%), supporting hypothesis H2a. It may be that as environmental dynamism increases (for example, with an increase in a standard deviation), the relationship between SIS and exploration innovation increases with the size of the interaction term, obtaining the path coefficient value of 0.590 (0.455 + 0.135) and representing a 23% increase in this relationship as shown in Figure 3.

Figure 3
Graph of the moderation effect of dynamism in the relationship between SIS and exploration innovation, according to recommendations by Dawson (2014).
Source: Prepared by the authors using SmartPLS v.3 software.

Our post-analysis study allowed us to identify and interpret potential ‘unobserved heterogeneity’ in the relationship between SIS and exploration and exploitation innovation, verifying the existence of factors not included in the original analysis and that can explain differences found between the various groups of companies. The final mix technique (FIMIX-PLS) was used as recommended by Hair et al. (2016) and Matthews, Sarstedt, Hair and Ringle (2016) in the PLS-PM analyses.

To identify the number of unobservable segments, the FIMIX-PLS algorithm (SmartPLS software v3) (Ringle, Wende, & Becker, 2015), was used and performed 10 times for g = 2-5 segments using the Akaike information criteria (AIC), modified AIC with factor 3 (AIC3), Bayesian information criterion (BIC), consistent AIC (CAIC), Hannan-Quinn criterion (HQ), and standardized entropy statistics (EN), which serve as satisfactory criteria for the selection of segments.

Indicators with lower values on certain information criteria indicate the best segment solution with an EN value above 0.50 (Hair et al., 2018). According to the authors, the first criterion determines whether a combination of AIC3 and CAIC values is lower by segment. As a second criterion, the lowest AIC3 and BIC set values should be considered. As a third criterion, the lowest values are found for the combination of AIC4 and BIC. If none of the previous criteria is met, the general recommendation is to choose the segment with the lowest value as indicated by the AIC and more segments than those indicated by the criterion of the MDL5.

In addition to these criteria, the minimum sample size must be met, which in this study was identified as 50 cases, that is, the maximum number of structural paths that point to the dependent variable (5) multiplied by 10 according to criteria used in the methodological literature (Hair et al., 2017). The criteria presented in Table 6 specify that the 2-segment solution is the most suitable, and thus, it was possible to proceed with the prediction-oriented segmentation approach (PLS-POS).

Table 6
Adjustment indices for a one to five segment solution.

Note. AIC - Akaike information criteria; AIC3 - AIC with modified factor 3; AIC4 - AIC with modified factor 4; BIC - Bayesian information criterion; CAIC - consistent AIC; HQ - Hannan-Quinn criterion; MDL5 - minimum description length with factor 5; EN - entropy statistics (normalized). Source: Elaborated by the authors.

Table 7
Relative segment size.

Note. Source: Elaborated by the authors.

To assess differences in the segments and alternative solutions, the PLS-POS algorithm (SmartPLS v3 software) (Ringle et al., 2015) was performed for the two segments proposed by FIMIX-PLS, and R² values were compared with the PLS-POS of the dependent variables (original R^2, R² ‘segment 1,’ R² ‘segment 2,’ and mean of the R² of the PLS-POS). Table 8 shows differences between the R² of the original sample and the other R² (‘segment 1,’ ‘segment 2,’ and the PLS-POS average), requiring complementary interpretations between the groups as recommended by the literature (Hair et al., 2018).

Table 8
PLS-POS solution by segment.

Note. Desirable ranges of metrics indicative of model quality = R² values of 0.67 (substantial), 0.33 (moderate), or 0.19 (weak) (Chin, 1998; Henseler et al., 2009). Source: Elaborated by the authors.

For the assessment of environmental dynamism in the post-analysis, we used the mean value of this variable to assign groups with low and high environmental uncertainty based on the procedure adopted in the SIS and dynamism study performed by Yayla and Hu (2012). Thus, values above the average of the respective dimension were considered to denote high dynamism (‘high dynamism,’ 147 cases), and values below the average were considered to denote low dynamism (‘low dynamism,’ 109 cases).

Table 9, in comparing cell counts, shows that the second segmentation (PLS-POS ‘segment 2’) served as the best combination for the variables of environmental dynamism, sector, and company size with an appropriate overlap of 65% found for the sample coverage as recommended in the literature (Hair et al., 2018).

Table 9
Cross tabulation by PLS-POS segment and moderation variables.

Note. Criteria used for the definition of ‘high dynamism’ and ‘low dynamism’ groups followed the procedure adopted by Yayla and Hu (2012). Source: Elaborated by the authors.

From the differences presented by the PLS-POS segment (Tables 8 and 9), the structural models for ‘segment 1’ and ‘segment 2’ were calculated. Table 10 presents the results of the structural models with statistical significance (p-value < 0.05), convergent validity, and reliability according to Hair et al. (2017). Additionally, the PLS-MGA method was used to compare the path coefficients of the models to those of the original sample, revealing differences in the relationship between SIS and exploitation innovation (| p1 - p2 | = 0.326; t-value = 6.185; p-value < 0.001) and in the relationship between SIS and exploration innovation (| p1 - p2 | = 0.284; t-value = 10.463; p-value < 0.001).

Table 10
Aggregated results by group.

Note. Desirable ranges of metrics indicative of model quality = p-value < 0.05 (Hair et al., 2017); R² values of 0.67 (substantial), 0.33 (moderate), or 0.19 (weak) (Chin, 1998; Henseler et al., 2009). Source: Elaborated by the authors.

This study extends the literature on information systems and strategy-as-practice by showing how SIS can be used as an alternative means for Brazilian organizations to integrate deliberative and emerging strategy-as-practice approaches to promote innovation in an environment of environmental uncertainty.

The study demonstrates that theoretical foundations of the strategy-as-practice approach, which can often be vague and abstract in conceptual studies, can be broken down into specific activities that can be measured with the incorporation of technological applications. In other words, the article has examined how IS applications are incorporated into strategic processes, whose aggregation and synergies encompass an organizational capacity to promote exploitation and exploration innovation. Thus, in exploring how IT can add value to strategy, especially in dynamic and turbulent environments (Kohli & Grover, 2008), the study identifies and examines how IS applications are fundamental to promoting strategy-as-practice (taking into account recent IS and strategy-as-practice studies) (Whittington, 2014) and to enabling innovation in the organization as mentioned by Marabelli and Galliers (2017).

Additionally, the study contributes to the literature on the strategy-as-practice approach by showing that the ubiquity of IT through an SIS incorporated into the strategic process enables innovation, breaking with the rigidity paradox of strategic planning not influencing innovation mentioned in studies of strategic planning and innovation (Arend et al., 2017; Song, Im, Van Der Bij, & Song, 2011). Thus, the study contributes to the literature on information systems and strategy-as-practice by revealing a viable alternative means through which companies can respond flexibly and quickly to the challenges of external dynamism and develop strategic initiatives of radical and incremental innovation.

In practice, the results of this study offer executives a clear understanding of how IS applications incorporated into strategic planning enable strategic-as-practice (deliberative and emerging strategies) in an organization to promote exploitation and exploration innovation. Thus, the study demonstrates the importance of companies investing in IS applications to promote the practice of strategy in an organization, leverage innovation, and face environmental uncertainty.

Therefore, the incorporation of IS applications into strategy activities will not only result in productivity and increase the speed of internal activities to enable incremental innovation but will also allow companies to engage in radical innovation in the face of market opportunities in new environments previously characterized by competition.

Additionally, from our post analysis results, it is evident that IS applications incorporated into the strategic process allow organizations (of different sectors and sizes and facing different levels of environmental uncertainty) to have the different stances on exploitation and exploration innovation under conditions of high dynamism.

Despite the methodological rigor adopted in this study in structuring the collection instrument, qualifying the respondents, and handling data, nonprobabilistic sampling for convenience is considered a limitation of this study, as it does not allow for generalizations.

Future research may investigate categories that characterize the heterogeneous subgroups identified in the post-analysis section; expand our investigation of the relationship between SIS and innovation by including new constructs that enable an organization to face the challenges of dynamism; investigate whether ambidextrous innovation (simultaneous exploration and exploitation innovation) can be enabled by SIS in conditions of dynamic environmental uncertainty; and develop longitudinal studies to identify the possible formation of SIS through innovation initiatives. These little explored areas can be investigated in studies to come.

* Corresponding Author